Compression of multi-year meteorological data

Dynamic simulation programes require hourly values of solar radiation and ambient temperature forming large files, which are usually difficult to handle with available personal computers (PC). This report describes stochastic models of these variables which have been constructed to overcome this difficulty. They are based on Markov chains and autoregressive processes, determined using multi-year hourly data of both variables. A validation of the model has been carried out for five different Swiss locations. It has shown that the main statistical characteristics of these variables are reproduced by the models. A very good agreement was also obtained between results of dynamic simulations carried out using measured and synthetic data. The generalization of the method to 30 Swiss locations has been made to facilitate the transfer of these developments into practice.